No abstract

No abstract

Recent advances in computer vision has led to a growth of interest in deploying visual analytics model on mobile devices. However, most mobile devices have limited computing power, which prohibits them from running large scale visual analytics neural networks. An emerging approach to solve this problem is to offload the computation of these neural networks to computing resources at an edge server. Efficient computation offloading requires optimizing the trade-off between multiple objectives including compressed data rate, analytics performance, and computation speed. In this work, we consider a "split computation" system to offload a part of the computation of the YOLO object detection model. We propose a learnable feature compression approach to compress the intermediate YOLO features with light-weight computation. We train the feature compression and decompression module together with the YOLO model to optimize the object detection accuracy under a rate constraint. Compared to baseline methods that apply either standard image compression or learned image compression at the mobile and perform image decompression and YOLO at the edge, the proposed system achieves higher detection accuracy at the low to medium rate range. Furthermore, the proposed system requires substantially lower computation time on the mobile device with CPU only.

Gold deposits in Precambrian cratons were mostly generated during the formation and stabilization of the cratons, but the North China craton is unusual in that its gold deposits were mainly formed ∼1.7 b.y. after its stabilization. A magmatic-hydrothermal origin or mantle-derived fluid source has been proposed for the giant gold deposits of the Jiaodong District in the eastern North China craton, but direct evidence is sparse, and the mineralization processes remain controversial. Here, we present the results of a comprehensive geological, geochronological, and geochemical study of the gold mineralized Xiejia diorite beneath the Linglong ore field at Jiaodong to link the gold mineralization to underlying magmatism. Magmatic zircon and titanite grains from the Xiejia diorite have laser ablation−inductively coupled plasma−mass spectrometry (LA-ICP-MS) U-Pb ages of 121.3 ± 0.9 Ma to 120.8 ± 1.1 Ma and 121.7 ± 3.9 Ma, respectively, which are indistinguishable from the time of gold deposition throughout the Jiaodong District as constrained by previous studies. The diorite has a shoshonitic composition and is characterized by strong enrichment in large ion lithophile elements (LILEs) and light rare earth elements (LREEs) along with significant depletion in high field strength elements (HFSEs) and heavy rare earth elements (HREEs). Samples of the diorite have high initial 87Sr/86Sr ratios, but low εNd(t) and ɛHf(t) values and low Pb isotope ratios. These geochemical characteristics are akin to those of contemporaneous mafic dikes in most gold mines at Jiaodong, indicating that the Xiejia diorite was most likely derived from an enriched lithospheric mantle source. The upper part of the diorite intrusion is pervasively altered and mineralized, containing an average of 0.32 g/t Au, but locally up to 7.59 g/t. Hydrothermal titanite from the mineralized diorite has a LA-ICP-MS U-Pb age of 122.3 ± 4.3 Ma, which is consistent with the gold-bearing pyrite Re-Os isochron age of 122.5 ± 6.7 Ma. Ore-related sericite aggregates from the Dongfeng gold deposit above the Xiejia diorite have a 40Ar/39Ar plateau age of 122.6 ± 1.3 Ma. Pyrite from the mineralized diorite yielded δ34SCDT values of 2.1‰−9.7‰, which are comparable with those of pyrite (δ34SCDT = 5.8‰−8.1‰, where CDT indicates the Canyon Diablo troilite standard) from gold ores of Dongfeng. Pyrite grains from both groups also have similar Pb isotope compositions. The S and Pb isotope data are consistent with values of mafic dikes that are spatially and temporally associated with gold veins in the Linglong ore field. The results presented here thus indicate a possible genetic link between gold mineralization in the Xiejia diorite and underlying magma presumably represented by the Xiejia diorite. The auriferous fluids exsolved from that magma and migrated upward along the Potouqing fault to form the Dongfeng gold deposit above the Xiejia diorite. The mineralized diorite thus links shallow gold mineralization to deep-seated mantle-derived magmatism generated during the extensive destruction of the North China craton induced by the rollback of the subducted paleo-Pacific plate.

Deep learning has made great strides for object detection in images, with popular models including Faster R-CNN, YOLO, and SSD. The detection accuracy and computational cost of object detection depend on the spatial resolution of an image, which may be constrained by both the camera and storage considerations. Furthermore, original images are often compressed and uploaded to a remote server for object detection. Compression is often achieved by reducing either spatial or amplitude resolution or, at times, both, both of which have well-known effects on performance. Detection accuracy also depends on the distance of the object of interest from the camera. Our work examines the impact of spatial and amplitude resolution, as well as object distance, on object detection accuracy and computational cost. As existing models are optimized for uncompressed (or lightly compressed) images over a narrow range of spatial resolution, we develop a resolution-adaptive variant of YOLOv5 (RA-YOLO), which varies the number of scales in the feature pyramid and detection head based on the spatial resolution of the input image. To train and evaluate this new method, we created a dataset of images with diverse spatial and amplitude resolutions by combining images from the TJU [1] and Eurocity [2] datasets and generating different resolutions by applying spatial resizing and compression. We first show that RA-YOLO achieves a good trade-off between detection accuracy and inference time over a large range of spatial resolutions. We then evaluate the impact of spatial and amplitude resolutions on object detection accuracy using the proposed RA-YOLO model. We demonstrate that the optimal spatial resolution that leads to the highest detection accuracy depends on the 'tolerated' image size (constrained by the available bandwidth or storage). We further assess the impact of the distance of an object to the camera on the detection accuracy and show that higher spatial resolution enables a greater detection range. These results provide important guidelines for choosing the image spatial resolution and compression settings predicated on available bandwidth, storage, desired inference time, and/or desired detection range, in practical applications.